A Multi-Sensor Data Fusion Method for Intelligent Aging Condition Identification of Viscoelastic Sandwich Structure

Abstract

Viscoelastic sandwich structure is playing an important role in mechanical equipment. The aging condition identification of viscoelastic sandwich structure is essential for monitoring structural health condition and guaranteeing the safe operation of mechanical equipment. The development of a reliable intelligent aging condition identification method is an ongoing attempt. By analyzing vibration response signals, a multi-sensor data fusion method using multi-scale permutation entropy (MPE) and ensemblebased incremental support vector machine (EISVM) is proposed in this paper. In this method, for obtaining complementary information, multiple sensors are mounted on the different locations of viscoelastic sandwich structure. For extracting effective feature information, MPE is introduced to reflect the structural aging condition change, and multiple feature sets of multiple sensors are obtained. With distance evaluation technique, multiple sensitive feature sets are respectively selected from the multiple original feature sets to discard irrelevant or redundant features. Based on ensemble learning and incremental learning, EISVM is developed to fuse all sensitive feature sets from different sensors for reliable intelligent classification. The proposed method is verified with a well-designed viscoelastic sandwich structure in which the viscoelastic layer subjected to accelerated aging. The testing results show the outstanding performance of the proposed method.